Z-Image Turbo | [image-to-image]
Tongyi-MAI's Z-Image Turbo delivers image-to-image generation with custom LoRA support at $0.0085 per megapixel. This 6B parameter model trades maximum resolution for rapid iteration speed and cost efficiency, making it 10-15x more cost-effective than alternatives with 117 generations per dollar. Built for developers running high-volume image transformation workflows where iteration speed and budget constraints matter more than absolute quality ceilings.
Built for: Product variant generation | Style transfer at scale | Rapid prototyping with custom LoRAs
At $0.0085 per megapixel versus $0.039+ for competitors, Z-Image Turbo delivers 117 generations per dollar, making it more cost-effective for high-volume workflows.
| Metric | Result | Context |
|---|
| Inference Steps | 1-8 steps | Configurable quality/speed tradeoff, default 8 steps |
| Cost per Megapixel | $0.0085 | 117 generations per $1.00 on fal |
| Batch Generation | 1-4 images | Per-request parallelization for variant testing |
| LoRA Capacity | Up to 3 LoRAs | Simultaneous application of custom training weights |
| Image Size Options | Auto-sizing | Adaptive resolution based on input dimensions |
Image-to-Image Generation With Custom Training
Z-Image Turbo processes reference images through Tongyi-MAI's 6B architecture, applying text prompts to transform existing visuals rather than generating from scratch. This contrasts with pure text-to-image models by preserving compositional structure while allowing targeted modifications through natural language.
What this means for you:
- Reference-guided generation: Transform existing images with text prompts while maintaining core composition and layout structure
- Custom LoRA support: Apply up to 3 trained LoRA weights simultaneously for style control, character consistency, or brand-specific aesthetics
- Strength control: Adjust transformation intensity from 0-1.0 to balance between original image preservation and prompt adherence
- Flexible output formats: Export as JPEG, PNG, or WebP with configurable quality settings for different distribution channels
Technical Specifications
| Spec | Details |
|---|
| Architecture | Z-Image Turbo 6B |
| Input Formats | Image URL (JPEG, PNG, WebP, GIF, AVIF) + text prompt |
| Output Formats | JPEG, PNG, WebP with configurable quality |
| Transformation Control | Strength parameter (0.0-1.0) for conditioning intensity |
| License | Commercial use permitted |
API Documentation | Quickstart Guide | Pricing
How It Stacks Up
FASHN Virtual Try-On V1.5 ($0.029 per image) vs Z-Image Turbo ($0.0085/MP): Z-Image Turbo prioritizes flexible transformation with custom LoRA support for general-purpose workflows at 3-4x lower cost. FASHN specializes in garment-specific try-on with body pose preservation for e-commerce product visualization.
Image Editing endpoints (Age Progression, Wojak Style, Reframe): Z-Image Turbo trades task-specific optimization for general-purpose flexibility through LoRA training. Specialized endpoints deliver preset transformations without requiring custom training, ideal for standardized editing workflows where consistency matters more than creative control.
fal-ai/z-image/turbo/image-to-image/lora
